Impact of ambient odors on mall shoppers’ emotions, cognition,and spendingA test of competitive causal theoriesJean-Charle...
ger, cardamom, licorice and chocolate are supposed toarouse a sense of romance, while rose combat depression(Hunter, 1995;...
2.3. Olfactory cues follow a different pathYet, the emotional route does not always emergeclearly. Spangenberg et al. (199...
al. (1984) do not challenge that cognition may be asufficient condition to produce emotions. ‘‘The question iswhether it i...
be positive. The first set of hypotheses is derived fromMehrabian and Russel (1974):Hypothesis 1a: A light and pleasing am...
The scent selection was based on Spangenberg et al.(1996) experimentation. They tested a series of 26 non-offensive odors ...
described in Table 2. Shopping mall perceptions werecaptured with a selection of Fisher’s (1974) semanticdifferentials (al...
Arousal influences pleasure (b = 0.575, t = 28.32). Pleas-ure mediates shoppers’ perceptions of the retail environment(b =...
The second set of hypotheses is partially supported: themediating effect of ambient scent on mood improves con-sumers’ per...
emotions. For instance, Chebat et al. (2001) showed thatstore music background has strong cognitive effects in termsof bot...
consumer choice on the service experience. J Consum Res 1991;18:174–84 (September).Hunter BT. The sales appeal of scents (...
Upcoming SlideShare
Loading in …5

Resourcd File


Published on

  • Be the first to comment

  • Be the first to like this

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide

Resourcd File

  1. 1. Impact of ambient odors on mall shoppers’ emotions, cognition,and spendingA test of competitive causal theoriesJean-Charles Chebat*, Richard MichonE´cole des Hautes E´tudes Commercials, 3000 chemin de la Coˆte Ste Catherine, Montreal, Quebec, Canada H3T 2A7AbstractThe authors tested the effect of ambient scents in a shopping mall environment. Two competing models were used. The first model isderived from the environmental psychology research stream by Mehrabian and Russel (1974) and Donovan and Rossiter (1982) whereatmospheric cues generate pleasure and arousal, and, in turn, an approach/avoidance behavior. The emotion–cognition model is supported byZajonc and Markus (1984). The second model to be tested is based on Lazarus’ (1991) cognitive theory of emotions. In this latter model,shoppers’ perceptions of the retail environment and product quality mediate the effects of ambient scent cues on emotions and spendingbehaviors. Positive affect is enhanced from shoppers’ evaluations. Using structural equation modeling the authors conclude that the cognitivetheory of emotions better explains the effect of ambient scent. Managerial implications are discussed.D 2003 Elsevier Science Inc. All rights reserved.Keywords: Store atmospherics; Scent; Emotion; Cognition; Expenses1. The ambient fragrance industry1.1. From air fresheners to aromatherapy, to environmentconditioningUntil recently, the fragrance industry was described as an‘‘embryonic industry’’ (Anonymous, 1998). Originally, itwas limited to air fresheners designed to remove unpleasantodors. Today, the home fragrance market is claiming relaxa-tion and revitalization benefits. Over the years, the Fra-grance Foundation says the industry has grown into a billiondollar business (Dunn, 1997). More and more products arescented, from sanitizing agents and toilet paper to toothpickand toothbrushes (Hunter, 1995). Coty introduced its Heal-ing Garden aromatherapy lines while Shiseido launched itsRelaxing fragrance (Butcher, 1998a).Drug and Cosmetic Industry, a trade publication,reports that the fragrance industry is also moving intothe conditioning of indoor environments using aromatechnology. Natural and chemical substances are releasedinto the ambient environment to improve feelings of well-being and even increase human performance. Envir-onmental conditioning systems are now found in homes,hotels, resorts, healthcare institutions, and retail stores.Aroma diffusion systems have been developed for theMirage, Treasure Island, Monte Carlo and Bally’s ResortHotels in Las Vegas, The Biscayne Bay Marriott andMarriott Airport Hotels in Miami. At Walt Disney Worldin Florida, the Magic House at Epcot Center includes a roomwith the fresh-baked smell of chocolate chip cookies toinduce feelings of relaxation and comfort (Butcher,1998a,b). Some in-house bakeries have been releasing syn-thetic scents to increase sales of bakery products. Dunkin’Donut recognized the importance of fresh coffee smell inattracting customers. Other companies such as Starbuckscoffee chains and Mrs. Field Cookies have followed suit(Hunter 1995).1.2. Myth or reality?How effective are the aromatherapy and environmentconditioning fragrances? It has been said that aromas fromlavender, basil, cinnamon and citrus flavor aid relax,whereas peppermint, thyme and rosemary invigorate. Gin-0148-2963/03/$ – see front matter D 2003 Elsevier Science Inc. All rights reserved.doi:10.1016/S0148-2963(01)00247-8* Corresponding author. Tel.: +1-514-340-6846; fax: +1-514-340-6432.E-mail address: (J.-C. Chebat).Journal of Business Research 56 (2003) 529–539
  2. 2. ger, cardamom, licorice and chocolate are supposed toarouse a sense of romance, while rose combat depression(Hunter, 1995; Amodio, 1998). For some industry repre-sentatives, the scientific evidence of aromatherapy may beless important than the ritualistic exercise based on feelingsrather than science (Butcher, 1998a,b). If the consumer buysthe concept of aromatherapy, he may get the effect he wantsregardless of what’s in the bottle. Much of the benefit ofaromatherapy may be people’s faith that it evokes (Amodio,1998). In the scientific literature, faith is replaced by theplacebo effect. Knasko et al. (1990) have shown that somepeople are emotionally and physically affected by theimaginary presence of an odor.The fragrance industry believes that the effect of scent onhumans is more than just folklore. It founded the OlfactoryResearch Fund, ‘‘a nonprofit charitable organization ded-icated to the study of the sense of smell and the positiveeffects of odor on human behavior’’ ( ORF financed at least partially several scientific uni-versity research projects on the fundamentals of odorperceptions; fragrance and psychology; olfactory condition-ing; scent and social behavior; and mood. The OlfactoryResearch Fund developed and ‘‘service marked’’ the con-cept of aromachology. The latter is said to be to aromather-apy what astronomy is to astrology.While there is a significant body of scientific researchon the effects of odors on human physiology and psy-chology, the research corpus on the effect of ambientscent on consumer behavior is much more limited (Morrinand Ratneshwar, 2000; Spangenberg et al., 1996; Gulas andBloch, 1995; Bone and Scholder, 1994). Retailers and serviceorganizations have not waited for consumer behaviorresearch before going ahead with environment conditioning.One is reminded of Paul Dukas’ Sorcerer’s Apprentice. Theuse of ambient scent in the retail environment can bebeneficial if congruent with the shopping environment.However, the same fragrance can become totally inefficient,or worst, have negative effects if used inappropriately. Ifambient scents are to be included in the retail marketingtoolbox, researchers and managers alike should try to under-stand how they work on consumers. This paper is one stepinto that direction.2. Ambient scent and marketing2.1. The processing of store atmospheric cuesIn the marketing literature, odors have been mentioned inpassing among the many retail/service atmospherics (Kotler,1973; Bitner, 1992, Baker et al., 1992, 1994). Somemarketing scholars have studied the effect of premisesclutter and cleanliness on consumers (Bitner, 1990; Garderand Siomkos, 1985), the effect of music (e.g., Yalch andSpangenberg, 1990; Dube´ et al., 1995), the effect of colors(Bellizzi et al., 1983; Crowley, 1993), the effect of lighting(Golden and Zimmerman, 1986), the effect of crowding(Eroglu and Machleit, 1990; Hui and Bateson, 1991; Erogluand Harrel, 1986) and the effect of ambient odor (Spangen-berg et al., 1996).Store atmospherics are a subset of the more generalresearch stream on the physical environment in servicebusinesses (Baker, 1986; Bitner, 1986; Booms and Bitner,1982). For an extensive and recent literature review, seeTurley and Milliman (2000). The retail environment hasbeen found to influence both shoppers and personnel’sbehavior, and to increase sales (Milliman 1982, 1986; Smithand Curnow, 1966; Stanley and Sewall 1976). Retail atmo-spherics can mediate product evaluation (Bitner 1986; Rap-poport 1982; Wheatly and Chiu 1977) and customersatisfaction (Bitner 1990; Harrell et al., 1980). Attitudestoward the store environment may be shown to be moreimportant in determining store choice than are attitudestoward the merchandise (Darden et al., 1983).2.2. The pleasure/arousal research tradition: the mediatingeffect of mood on cognition and behaviorMost marketing scholars studying retail atmosphericsfollowed Mehrabian and Russell’s pleasure/arousal researchtradition in which mood is a mediating factor betweenenvironmental cues and behavior. Mehrabian and Russel(1974) examined ambient (lighting and music) and socialcues (number and friendliness of employees) on respond-ents’ pleasure and arousal, and willingness to buy. Envir-onmental psychologists propose that individuals react totheir environment with two contrasting forms of behavior:approach and avoidance (Mehrabian and Russel, 1974),where approach is a desire to stay, explore and affiliate,and avoidance is the opposite. Much of the research on storeatmospherics has presumed a mediating effect of mood onconsumers’ cognition and behavior. Obermiller and Bitner(1984) showed that respondents who viewed retail productsin an emotionally pleasing environment evaluated productsmore positively than subjects who viewed the same productsin an unpleasant environment.Hui and Bateson (1991) found that in a crowdedenvironment, enhanced perceptions of personal controlwere related to increased pleasure. Negative affect associ-ated with crowding increases consumers’ desire to leave(Eroglu and Machleit, 1990). Positive affect encouragesshoppers to stay longer and interact with other employees(Babin and Darden, 1995; Dawson et al., 1990; Donovanand Rossiter, 1982; Hui and Bateson, 1991). Positive affectsimplifies consumers’ decision-making style (Babin et al.,1992), builds a positive store image (Darden and Babin,1994) and improves merchandise and service qualityperceptions (Baker et al., 1994). Positive affect inducedby background music stimulates the desire to affiliate(Dube´ et al., 1995). Although a number of studies focuson store atmospherics, the process through which cues arechanneled remains unclear.J.-C. Chebat, R. Michon / Journal of Business Research 56 (2003) 529–539530
  3. 3. 2.3. Olfactory cues follow a different pathYet, the emotional route does not always emergeclearly. Spangenberg et al. (1996) who studied howambient scent influenced store and product evaluationsfound no main or interactive effects regarding scent onmood. Researching the impact of ambient scent onevaluation and memory for familiar and unfamiliarbrands, Morrin and Ratneshwar (2000) concluded thatambient odors did not affect subjects’ mood or arousallevels. This suggests that olfactory cues may be pro-cessed through an alternate cognitive path. Consumerscommonly look for cues about the firm’s capabilities andquality (Berry and Clark, 1986; Shostack, 1977). Thesecues are found in the immediate physical firm’s envir-onment (Rappoport, 1982).3. Effects of odors on mood and cognition3.1. Pleasant or unpleasant odorsOdors appear difficult to identify. They are thus signific-antly different from visual and audio cues. (Schab, 1991).Consumers must depend on neighboring cues to identifyodors (Davis, 1981): for example, individuals are morelikely to recognize a lemon-scented product if presented ina yellow container (Ellen and Bone 1998). Odors areprimarily perceived in terms of their pleasantness orunpleasantness (Ehrlichman and Halpern, 1988; Buck andAxel, 1991). The affective dimension is commonly found inodor perceptions (Engen, 1982; Moskowitz, 1979; Schiff-man, 1979). One possible reason for this is that odors enterthe limbic system, i.e., that part of the brain at the center ofemotions (Spangenberg et al., 1996).3.2. Arousing odorsOdors also have the capacity to induce arousal. Lorig andSchwartz (1988) mention that the effects of odors areobserved by electroencephalographs (EEG). In the case ofodors, the relationship between arousal and pleasantness isnot linear: as odors get more intense, reactions tend tobecome more negative (Richardson and Zucco, 1989). Evenif they have not been tested for odors, olfactory stimulishould follow an inverted U-shape function (Spangenberg etal., 1996). However, the affective response to odors shouldnot obliterate the arguments in favor of the cognitive effectsof odors (Gulas and Bloch, 1995).3.3. Effects of odors on cognitive processingThe link between odors and cognition dates back to 1932,when Laird investigated how scented hosiery influencedwomen’s perceptions of quality. Scents that are congruentwith specific product actually improved product evaluation(Bone and Jantrania, 1992). Mitchell et al. (1995) conductedan experiment in computer-aided product selection in scen-ted rooms. Product congruent odors influenced informationprocessing. Spangenberg et al. (1996) tested if an ambient,nonoffensive scent affects store and product evaluations.Odors are often associated with objects, events andpersons. They stir up happy or sad memories. Perhaps, themost famous example comes from French novelist MarcelProust’s Remembrance of Things Past. In this monumentalwork, sensory cues arouse pleasant or melancholic feelingsby retrieving deeply entrenched, if not almost forgotten,memories. People’s reactions to odors may be tied to evokedassociations (Kirk-Smith, 1994).Is mood a necessary mediator of cognition? Bone andScholder (1998) report several research papers in whichodor effect may occur in the absence of a mood shift:(Cann and Ross, 1989; Ehrlichman and Halpern, 1988;Knasko, 1992; Spangenberg et al., 1996). They concludethat the consumer does not experience a mood shift, butsimply transfer the pleasantness/unpleasantness of the scentto the object.3.4. The interplay between mood and cognitionFew empirical studies on store atmospherics include bothaffect and cognition as mediators to shoppers’ behavior.Retail atmospheric research focused on the effect of envir-onmental cues on mood and approach avoidance. The waycues are processed through the maze of emotion/cognitionor cognition/emotion remains unclear. In the field of ambi-ent scent, studies combining emotion and cognition failed toshow a mood shift (Spangenberg et al., 1996; Morrin andRatneshwar, 2000). There is a definite need to furtherunderstand the interplay and the hierarchy of cognitionand emotion in the study of environmental cues.Potential managerial implications are important. Shouldretailers try to modify shoppers’ mood? Should they con-centrate their effort on shoppers’ perceptions of the storeenvironment or on product quality? One is easier done thanthe other. The dichotomy is between entertainment andinformation. Are mood shifts caused by entertainmentefforts or by informational cues? Two schools of thoughtmay contribute to our comprehension of these relationships.On one hand, there is the emotion–cognition approach. TheServicescapes model (Bitner, 1992) was constructed on theemotion–cognition theory. Zajonc and Markus (1984) con-tend that emotion can take place without antecedent cognit-ive processes. They argue that an emotion can be generatedby biological, sensory or cognitive events. Arousal andmotor activities are the hard representations of emotions.The experience of emotion, that requires a cognitive input,is the soft representation of affect. Only arousal is anecessary consequence of the generation of emotion. Thecognitive experience is not required to be part of theemotion process. For Zajonc and Markus, ‘‘the experienceof emotion is simply the cognition of having one.’’ Izard etJ.-C. Chebat, R. Michon / Journal of Business Research 56 (2003) 529–539 531
  4. 4. al. (1984) do not challenge that cognition may be asufficient condition to produce emotions. ‘‘The question iswhether it is a necessary cause’’ (p. 5).On the other hand, the cognition–emotion school ofthought (Lazarus, 1991) posits the causal role of cognitionas a necessary but not sufficient condition to elicit emotions.External and internal cues must be appraised in terms ofone’s own experience and goals. ‘‘Appraisal of the signific-ance of the person–environment relationship, therefore, isboth necessary and sufficient; without a personal appraisal(i.e., of harm or benefit) there will be no emotion; whensuch an appraisal is made, an emotion of some kind isinevitable’’ (p. 177).Two competing models combining emotions and cog-nition are being tested. In the first model, olfactory ambientcues stimulate positive emotions influencing shoppers’perceptions of their environment and product quality. Thealternate model hypothesizes that ambient scents mediatethe perceptions of the shopping environment and productquality, thus enhancing shoppers’ mood.4. Hypotheses and model buildingFigs. 1 and 2 present the two competing models withprimary and secondary path effects. In the first model (seeFig. 1), it is assumed that ambient scent influences consum-ers’ mood through the arousal and pleasure dimensions.Russell and Pratt (1980) found that pleasure and arousalwere independent dimensions. Berlyne (1971, 1974)hypothesized that arousal influences pleasure. The pathfrom arousal to pleasure is verified in a number of market-ing studies (Wakefield and Baker, 1998; Babin and Attaway,2000; Eroglu et al., 1998).Pleasant feelings are not necessarily correlated withstrong arousal (Dube´ et al., 1995; Spangenberg et al.,1998; Richardson and Zucco, 1989). Therefore, the primarypath between scent and mood may transit by both affectivedimensions. Considering the inverted U-shape effect ofscent intensity (Richardson and Zucco, 1989), the influenceof arousal over pleasure may either be positive or negative.Assuming a light pleasant scent, the effect of arousal shouldFig. 1. Proposed model for Hypotheses 1a–c and 2a–c. (Arrows represent primary paths.)Fig. 2. Alternate model for Hypotheses 3a and b and 4a and b. (Arrows represent primary paths.)J.-C. Chebat, R. Michon / Journal of Business Research 56 (2003) 529–539532
  5. 5. be positive. The first set of hypotheses is derived fromMehrabian and Russel (1974):Hypothesis 1a: A light and pleasing ambient scentarouses consumers.Hypothesis 1b: A light and pleasing ambient scentincreases consumers’ positive affect.Hypothesis 1c: Arousal induced by a light and pleasingambient scent should positively influence consumers’mood.The positive affect is expected to provoke a favorableperception of the shopping environment, under theapproach/avoidance model (Mehrabian and Russel, 1974;Donovan and Rossiter, 1982). There is also theoreticalsupport for linking pleasure with the perception of productquality, through transfer (Obermiller and Bitner, 1984).Hypothesis 2a: Consumers’ mood improves perceptionsof the shopping environment.Hypothesis 2b: Consumers’ mood improves perceptionsof product quality.Hypothesis 2c: Consumers’ perceptions of the shoppingenvironment impact the perception of product quality.The third hypothesis is congruent with the approach/avoidance model (e.g., Mehrabian and Russel, 1974) andfindings by Donovan and Rossiter (1982). It is believed thatconsumers who are in a favorable mood are more likely towant to affiliate with others, stay longer in the stores, andmake more purchases.Hypothesis 3a: A positive perception of the shoppingenvironment influences consumer spending.Hypothesis 3b: A positive perception of product qualityinfluences consumer spending.Fig. 1 underscores the interplay between research hypo-theses. Through this network of influences, our intention isto stress the mediating effect of ambient scent on mood, onperceptions and on behavior. In the first model, mood(affect) is an antecedent to perception (cognition).In the competing model (see Fig. 2), ambient scent isperceived by consumers without a mood shift (Bone andScholder, 1998; Morrin and Ratneshwar, 2000; Spangen-berg et al., 1996; Knasko, 1992; Ehrlichman and Halpern,1988). Consumers use ambient scent as an environmentalcue that impacts on product evaluation (Bitner, 1986, 1990;Rappoport, 1982).The fourth set of hypotheses is based on the envir-onmental cue theory (Berry and Clark, 1986; Rappoport,1982; Shostack, 1977), and is in line with the findings ofSpangenberg et al. (1996):Hypothesis 4a: A light and pleasing ambient scentdirectly affects consumers’ perception of the shoppingenvironment.Hypothesis 4b: A light and pleasing ambient scentinfluences consumer’s perception of product quality.The competing model presupposes that the perceptionsof the retail environment and of product quality are ante-cedents to consumers’ affect. Marketing scholars such asBagozzi and Moore (1994) and Bagozzi et al. (1999) haverelied on the cognitive theory of emotions to explainconsumers’ behaviors. The fifth hypothesis is derived fromthis literature and the findings by Baker et al. (1994).Hypothesis 5: Consumers’ perceptions of the retailenvironment and of product quality foster a morefavorable shopping mood resulting in more moneybeing spent.The first two hypotheses posit that affect is an antecedent tocognition (e.g., Zajonc and Markus, 1984), whereas the lasttwo assume that cognitive processes influence mood (e.g.,Lazarus, 1991).5. MethodThe proposed models were tested in a mall intercept witha two-factor experiment. Until now, other ambient scentexperiments had not been conducted on actual retail loca-tion. For instance, the study by Morrin and Ratneshwar(2000) was conducted in a laboratory, and the study bySpangenberg et al. (1996) in a simulated store. In our case,the experiment was conducted in a community shoppingmall located on the western outskirts of the Montrealmetropolitan area. The shopping mall is owned and operatedby one of Canada’s largest commercial property developer.Data were collected in two waves.The experiments were conducted in two consecutiveweeks: the last week of February 1998 for the control groupand the first week of March 1998 for the experimentalgroup. These two weeks are known by the shopping mallowners to be identical in terms of sales volume and shopperstraffic. Special care was taken by the mall director to cancelall special promotions by the retailers during the two weeksof the experiment.In the control wave, the shopping mall ambient olfac-tory atmosphere was not modified. There were no aggress-ive exogenous odors emanating say from food outlets orfragrance stores. During the second wave, a light pleasingscent was vaporized in the mall’s main corridor. Theambient scent was diffused in the shopping mall’s maincorridor located between two major retailers. Some 10diffusers released a citrus scent for 3 s every six min,thus, maintaining a continuous scent intensity. Specialcare was taken to ensure that the odor intensity reachedperceptual thresholds without bothering people. Exposedsubjects and controls had similar response patterns aboutthe pleasing [ F(1,568) = 0.267, P=.601] or the unpleasing[ F(1,565) = 2.103, P=.148] ambient odors in the shoppingmall. However, exposed subjects were more likely tobelieve the odors in the shopping mall were appropriate[ F(1,560) = 4.914, P=.027].J.-C. Chebat, R. Michon / Journal of Business Research 56 (2003) 529–539 533
  6. 6. The scent selection was based on Spangenberg et al.(1996) experimentation. They tested a series of 26 non-offensive odors on the affective and activation (arousal)scales originally developed by Fisher’s (1974) and usedby Crowley (1993) in environmental research. Scentcategories included floral, spice, wood, citrus and mint.The affective dimension comprised five items (positive,attractive, relaxed, and good). The activation scale (i.e.,arousal) was also made up of five attributes (stimulating,lively, bright, motivating and interesting). Ginger andlavender were identified as effectively neutral, whileorange and mint were found effectively pleasing.For the mall intercept experiment, a citrus scent (acombination of orange, lemon and grape) was chosen. Thescent category scored well with Spangenberg et al. (1996)pretest, and was also available from vendors. Citrus issignificantly different from lemon. People readily associatethe smell of lemon with cleaning products (Bone andJantrania, 1992). Furthermore, the scent was not congruentwith any specific products sold in the shopping center, asrecommended by Spangenberg, Crowley and Henderson.Graduate marketing students, who were asked not towear perfume, were responsible for the administration ofquestionnaires. Sampled individuals were not aware of theresearch objectives. They were simply invited to fill-in aself-administered questionnaire on their shopping trip. Datacollection covered all weekdays and day parts for adequaterepresentation. Some 145 subjects exposed to the scenttreatment completed the questionnaire. The control groupincluded 447 individuals. (The larger size of the controlgroup was required for other experiments).The experimental sample contains marginally morewomen than normally anticipated. Yet, the chi-square stat-istics are reassuring. Randomly selected participants in thecontrol and the experimental groups had similar sociodemo-graphic profiles (Table 1). The distribution of participants issomewhat similar to that in the Province of Quebec. Themedian age of the Quebec adult population is 43 years whilethat of the control and the experimental sample is 42 and 43years old, respectively. The median household income is47,000 dollars for the control group, 43,000 for the experi-mental group and 47,000 dollars for the general population.There are no concentrations of young or old participantswith different olfactory sensitivities.Subjects and controls were probed on their perceptionsof product quality, using the three-item scale developed byBellizi et al. (1983). The three product quality scale itemshave a Cronbach’s alpha coefficient of .87. Scale items areTable 1Sociodemographic profiles of control and experimental groupsSample size Control group, N = 447 (%) Experimental group, N = 145 (%)Gender (c2= 2.99, df = 1, P=.10) Females 55 63Males 45 37Age (c2= 7.32, df = 5, P=.20) 18–24 10 1325–34 19 2135–44 29 2245–54 23 1955–65 11 1765 + 8 8Education (c2= 1.37, df = 3, P=.71) Primary 3 1High school 33 31College 35 35University 29 33Household income (c2= 5.22, df = 8, P=.73) Under US$30,000 24 28US$30,000–49,000 29 31US$50,000–79,000 25 26US$80,000 + 13 9Refusals 8 6Language (c2= 0.44, df = 1, P=.60) French 83 85English 17 15Table 2Scale items, with alpha coefficients and factor loadingsItems Factor 1 Factor 2 Factor 3 Factor 4Product quality (a=.87; Bellizi et al., 1983)Outdated/up to date À .040 .292 .831 .031Inadequate/adequate À .095 .266 .870 À .000Low/high quality À .071 .290 .816 À .020Shopping environment (a=.90; Fisher, 1974)Boring/stimulating À .084 .873 .224 À .069Unlively/lively À .074 .828 .201 À .119Uninteresting/interesting À .050 .835 .274 À .079Pleasure (a=.96; Mehrabian and Russel, 1974)Unhappy/happy .926 À .114 À .069 À .045Annoyed/pleased .923 À .134 À .059 .024Unsatisfied/satisfied .916 À .137 À .055 À .023Melancholic/contended .905 À .122 À .033 .018Arousal (a=.76; Mehrabian and Russel, 1974)Relaxed/stimulated .579 À .129 À .049 .630Calm/excited .232 À .098 À .017 .847The alpha coefficients are for the selected items only. The factorloadings refer to the full scale, not just selected items.J.-C. Chebat, R. Michon / Journal of Business Research 56 (2003) 529–539534
  7. 7. described in Table 2. Shopping mall perceptions werecaptured with a selection of Fisher’s (1974) semanticdifferentials (alpha coefficients=.94). Mehrabian and Rus-sel’s (1974) pleasure and arousal items were also adminis-tered (alpha coefficient=.91). Participants were asked howmuch they had spent, excluding groceries, during theirshopping trip.To illustrate the cognitive and affective paths of ambientscent, structural equation modeling (SEM) was used (EQSfor Windows 5.7a). Variables described in Table 2 wereentered in both models. All efforts were made to keep thestructural model to a manageable size. Only the mostrelevant items from Fisher’s (1974) semantic differentials,and from the Mehrabian and Russel’s (1974) pleasure/arousal scale, were entered in the model.Exploratory factor analyses helped select items with thehighest factor loading on one dimension. Both models to betested were built with three indicators per latent variable(Bollen, 1989, p. 244), with the exception of Arousal (twoindicators). Bollen (1989) underlines that the two-indicatorrule is a sufficient condition for measurement models withmore than one x variable (p. 244). One of the Arousalindicators also loads high on the Pleasure dimension,showing a correlation between the two nonorthogonal con-structs (Crowley, 1993; Berlyne, 1971, 1974).Confirmatory factor analysis (CFA) performed on thefactor structure in Table 2 substantiates the exploratoryanalysis findings (Satorra–Bentler scaled c2= 52.24,df = 46; P=.244; CFI = 0.999). As anticipated, the percep-tions of product quality and of the shopping mall envir-onment are correlated; so are consumers’ pleasure andarousal effects. However, the two pairs of constructs areorthogonal to each other.Consumer spending has only one indicator, actual dollarsdisbursements during the shopping trip, excluding groceries.Its error variance has been set to zero. The presence or theabsence of scent is represented by a dummy variable (1 and0). Bagozzi (1994) and Bagozzi and Yi (1989) have useddummy variables with structural equation models in experi-mental designs. Because of the categorical variables, themodels were estimated with Yuan–Bentler corrected AGLSchi-square statistics, an asymptotically distribution-free(ADF) statistic added to EQS 5.7a (January 1998).6. Research findings6.1. Model 1In the first model, where emotions are antecedent tocognition, the effect of the ambient scent on shoppers’ moodcancels itself through Arousal (b = 0.063, t = 5.28) andPleasure (b = À 0.025, t = À 2.49). The odor may be some-what arousing but fails to directly induce pleasure. Thecombined effect of scent on shoppers’ mood is almost null(b: 0.063 Â 0.575 À 0.025 = 0.009). Instead the LagrangeMultiplier test suggests direct paths between ambient scentand the cognition variables (mall perception: b = 0.099,t = 13.68; perception of product quality: b = 0.036,t = 4.91). These paths have been hypothesized in the com-peting model. Fig. 3 shows the standardized parameterpaths. The structural parameter estimates, and t values arefound in Table 3.Fig. 3. Model 1 — affect–cognition model: emotion as an antecedent to cognition (standardized parameters).Table 3Model 1 — affect–cognition structural model estimates (t values)Pleasure = Arousal À Ambient Scent0.575 (28.32) 0.025 ( À 2.49)Arousal = Ambient Scent0.063 (5.28)Environment = Pleasure + Ambient Scent0.158 (34.28) 0.099 (13.68)Quality = Pleasure + Environment + Ambient Scent0.009 (2.66) 0.445 (28.35) 0.036 (4.91)Spending = Environment + Quality0.089 (3.05) 0.062 (1.54)J.-C. Chebat, R. Michon / Journal of Business Research 56 (2003) 529–539 535
  8. 8. Arousal influences pleasure (b = 0.575, t = 28.32). Pleas-ure mediates shoppers’ perceptions of the retail environment(b = 0.158, t = 34.28). However, this positive mood is notdue to ambient scent. As expected, the retail atmosphericsmediate the perception of product quality (b = 0.445,t = 28.35). The effect of pleasure on the perception ofproduct quality is marginally significant (b = 0.009,t = 2.66). In this emotion–cognition model, the perceptionsof the retail environment (b = 0.089, t = 3.05) and of productquality (b = 0.062, t = 1.54) have little effects on shoppers’spending. The path between product quality and consumerspending is not significant. Fit statistics associated with thismodel are perfectible (Yuan–Bentler AGLS c2= 90.18,df = 63; P=.014, RMSEA= 0.034; CFI = 1.00).6.2. Model 2The competing model (Fig. 4) assumes that perceptioninfluences shoppers’ mood. This model yields a better fit(Yuan – Bentler AGLS c2= 36.52, df = 63, P=.997,RMSEA= 0.00; CFI = 1.00) (see Table 4). The effect ofambient scent on shoppers’ perceptions is nonambiguous.The presence of odor primarily influences the perception ofthe shopping environment (b = 0.106, t = 15.81) and that ofproduct quality (both directly: b = 0.043, t = 4.86) and indi-rectly: 0.106 Â 0.551). The perception of mall atmosphericsmediates shoppers’ arousal (b = 0.396, t = 11.00). The per-ception of product quality influences pleasure (b = 0.104,t = 5.27), although it is somewhat moderated through itsnonarousing effect (b = À 0.163, t = À 4.21). As anticipated,arousal stimulates pleasure (b = 0.358, t = 27.63). Yet, shop-pers’ emotions are not a strong antecedent to consumerspending. The model underscores that shoppers’ spending ismediated primarily by perceptions of product quality(b = 0.126, t = 14.48), not by mood (b = 0.026, t = 4.43).7. DiscussionThe two models under investigation are built from thesame latent variables and indicators. They only differ on theorder of mediating variables. The first model follows theenvironment psychology paradigm where positive (nega-tive) affects stimulate approach (avoidance) behaviors. Fitstatistics are marginal. Despite a strong CFI (1.00) and lowRMSEA (0.034), the Yuan–Bentler AGLS chi-square fitspoorly (c2= 90.18, df = 63, P=.014). The structural equationmodel supports Hypothesis 1a (a light and pleasing ambientscent arouses consumers) but rejects Hypothesis 1b (a lightand pleasing ambient scent increases consumers’ positiveaffect). The direct and indirect effects of ambient scent overpleasure cancel each other. The path from ambient scent toarousal (b = 0.063, t = 5.28) is significant, indicating that thepresence of odor may have an arousing effect. At the sametime, the path from ambient scent to pleasure has a negativecoefficient (b = À 0.025, t = À 2.49), suggesting an attenu-ating effect of scent over pleasure. EQS’s Lagrange Mul-tiplier (LM) test keeps suggesting more robust paths fromambient scent to perception variables (b = 0.099, t = 13.68for the mall environment; and b = 0.036, t = 4.91 for productquality). This finding is in opposition with one of the mainelements of the approach/avoidance model: pleasure andarousal do not mediate the effects of environmental cues onperceptions and behaviors.Fig. 4. Model 2 — cognition–affect model: cognition as antecedent to emotion (standardized parameters).Table 4Model 2 — cognition–affect model estimates (t values)Environment = Ambient Scent0.106 (15.81)Quality = Environment + Ambient Scent0.551 (27.75) 0.043 (4.86)Arousal = Environment À Quality0.396 (11.00) 0.163 ( À 4.21)Pleasure = Arousal + Quality0.358 (27.63) 0.104 (5.27)Spending = Pleasure + Quality0.026 (4.43) 0.126 (14.48)J.-C. Chebat, R. Michon / Journal of Business Research 56 (2003) 529–539536
  9. 9. The second set of hypotheses is partially supported: themediating effect of ambient scent on mood improves con-sumers’ perception of their shopping environment (Hypo-thesis 2a) and of product quality (Hypothesis 2b). The effectof ambient scent on shoppers’ mood has already beenshown as insignificant. Consumers’ mood (pleasure) —not induced by ambient scent — affects the perception ofthe shopping atmospherics (b = 0.158, t = 34.28). Asexpected in the approach/avoidance theory, the moodenhanced perception of the retail environment influencesshoppers’ spending (b = 0.089, t = 3.05). Yet, the perceptionof product quality, in this model, has no significant effect(b = 0.062, t = 1.54) on consumer spending. Hypothesis 3a isaccepted, while Hypothesis 3b is rejected.The alternate model is more robust than the previous one:the Yuan–Bentler AGLS chi-square statistic provides strongindications of the model fit (c2= 36.52, df = 63, P=.997).The SEM depicting perceptions as an antecedent to emo-tions supports the forth set of hypotheses: a light andpleasing ambient scent directly affects consumers’ percep-tion of the shopping environment (Hypothesis 4a) and ofproduct quality (Hypothesis 4b). The paths from ambientscent to the perception of the retail environment (b = 0.106,t = 15.91) and to that of product quality (b = 0.043, t = 4.86)are both significant. However, the cognitive effect of ambi-ent scent primarily passes through the perception of theretail environment.Consumers’ spending is more likely to be induced bycognitive processes than by mood alone. The first part ofHypothesis 5 is strongly supported, while the second part ofthe proposition must be regarded with circumspection (Themediating effect of ambient scent on the retail environmentmodifies the shoppers’ perception of quality which, in turn,induces a more favorable shopping mood resulting in moremoney being spent.). The path coefficients from ambientscent to the perception of the retail environment (b = 0.106,t = 15.81) and from the retail environment to the perceptionof product quality (b = 0.551, t = 15.81) are convincing.Shoppers’ perceptions of their environment and of productquality affect their mood. The paths from the perception ofthe retail environment to arousal (b = 0.396, t = 11.0) andfrom the perception of product quality to pleasure(b = 0.104, t = 5.27) cannot be rejected. The perception ofproduct quality does not have an arousing effect(b = À 0.163, t = À 4.21). Consumer spending is primarilyinfluenced by the perception of product quality (b = 0.126,t = 14.48) rather than by mood (b = 0.026, t = 4.43).8. ConclusionAmbient scent contributes to the building of a favorableperception of the mall environment, and indirectly ofproduct quality. Although significant, product quality maybe viewed as a necessary but not sufficient condition toexplain increased shopper’s spending. Nonenvironmentalfactors mediate consumer spending. For example, Babinand Attaway (2000) have shown that shoppers’ utilitarianand hedonic affect behaviors.Retailers should seriously consider ambient scent in theirmarketing toolbox. It is probably among one of the leastexpensive techniques to enhance shoppers perceptions.Product-related or congruent scent may be effective toincrease the sales of a particular product (Bone and Jan-trania, 1992), but may also jeopardize the sales of otherproducts (Spangenberg et al., 1996). Effective ambient scentshould support all products in the store (e.g., office suppliesand furniture; Gulas and Bloch, 1995).In this experiment, the environmental cue (ambient scent)directly affects shoppers’ perceptions. The latter have asignificant influence over consumers’ mood. The best fittingmodel supports the cognitive theory of emotions. Bothcognitive and affective dimensions are central to market-oriented managers. For retailers, whether emotions precedecognition or vice versa, the question may appear trivial. Yet,one does not go without the other. It may be even moretrivial to try to enhance mood, for example, through someform of entertainment, if it does not influence shoppers’perceptions of the retail environment and that of product/service quality.Our findings strongly support the model derived fromLazarus (1991), which contradicts one of the basic tenets ofthe approach/avoidance model, that is the mediating effectsof consumers’ mood in the atmosphere–perception relation-ship. In this specific case, odors do impact significantly onperceptions of both product quality and shopping envir-onment. Mood (i.e., pleasure and arousal) contributes verylittle (although the contribution is significant) on spending.Odors affect the perception of product both directly andthrough the perceptions of the mall environment. In addi-tion, the effect of mall perception on product quality is verystrong. This result is central to store atmospherics: theshopping mall plays the role of a global packaging for theproducts sold. Since odors may affect the mall perception,retailers should consider this specific odor study here, i.e.,citrus as a powerful way of influencing product perception.This is even more important since product perceptions affectsignificantly sales (b = 0.126; t = 14.48). Perception of themall environment affects shoppers’ arousal (b = 0.396;t = 11.00) very strongly, whereas perception of productquality has very little impact (although it is significant) onemotions. Perception of mall environment cannot be down-played: it impacts on the perception of product, which, inturn, affects spending. It also impacts on emotions but theydo not contribute much to spending.We conclude from this that the importance granted toemotions in the store atmospheric literature may have beenoverstated. In other words, the approach/avoidance modelthat strongly stressed the emotional effects of the envir-onment to the detriment of the study of the meaning of theenvironment. Odors, as well as music, may mean somethingto the consumers more powerfully than they affect theirJ.-C. Chebat, R. Michon / Journal of Business Research 56 (2003) 529–539 537
  10. 10. emotions. For instance, Chebat et al. (2001) showed thatstore music background has strong cognitive effects in termsof both cognitive response and information processing.They explain that music background may stimulate (orcancel) strong cognitive processes. We reason that odorsmay have similar effects. Odors may affect the cognitiveprocesses (as shown by Laird, 1932; Mitchell et al., 1995,for instance). We then strongly suggest to reorient, at leastpartially, the store atmospherics toward the understanding ofambient meaning to shoppers.AcknowledgmentsThe authors gratefully acknowledge the vital cooperationof Ivanhoe and a research grant from FCAR.ReferencesAmodio JV. Mood makers: an in-depth look at how scents claim to boostyour spirit and de-stress your nerves. Good Housekeeping 1998;227(5):53–5 (November).Anonymous. Home sweet home. Eur Cosmet Mark 1998:519 (December).Babin BJ, Darden WR. Consumer self-regulation in a retail environment. JRetailing 1995;71:47–70 (Spring).Babin BJ, Attaway JS. Atmospheric affect as a tool for creating value andgaining share of customer. J Bus Res 2000;49:91–9 (August).Babin BJ, Darden WR, Griffin M. Work and/or fun: measuring hedonic andutilitarian shopping value. J Consum Res 1992;20:644–56 (March).Bagozzi RP. Advanced methods in marketing research. Cambridge (MA):Basil Blackwell, 1994.Bagozzi RP, Yi Y. On the use of structural equation models in experimentaldesigns. J Mark Res 1989;27:271–84 (August).Bagozzi RP, Moore DP. Public service advertisement: emotion and empathyguide prosocial behavior. J Mark 1994;58:56–70.Bagozzi RP, Gopinath M, Nyer PU. The role of emotions in marketing. JAcad Mark Sci 1999;27(2):184–206.Baker J. The role of the environment in marketing services: the consumerperspective. In: Czepiel JA, Congram CA, Shanaham J, editors. Theservices challenge: integrating for competitive advantage. Chicago (IL):American Marketing Association, 1986. p. 79–84.Baker J, Levy M, Grewal D. An experimental approach to making retailstore environmental decisions. J Retailing 1992;68(4):445–60 (Winter).Baker J, Grewal D, Parasuraman A. The influence of store environment onquality inferences and store image. J Acad Mark Sci 1994;22(4):328–39.Bellizzi JA, Crowley AE, Hasty RW. The effects of color in store design. JRetailing 1983;59:21–45.Berlyne DE. Aesthetics and psychobiology. New York: Wiley, 1971.Berlyne DE. Studies in the new experimental aesthetics New York:Wiley, 1974.Berry L, Clark T. Four ways to make services more tangible. Business1986;36(4):53–5 (October–December).Bitner MJ. Consumer responses to the physical environment in servicesetting, creativity in services marketing. In: Venkatesan M, Schmalen-see DM, Marshall C, editors. Chicago: American Marketing Associa-tion, 1986. p. 89–93.Bitner MJ. Evaluating service encounter: the effects of physical surround-ing and employee responses. J Mark 1990;54(2):69–82.Bitner MJ. Servicescapes: the impact of physical surroundings on custom-ers and employees. J Mark 1992;56(2):57 (April).Bollen KA. Structural equations with latent variables. New York: Wiley,1989.Bone PF, Jantrania S. Olfaction as a cue for product quality. Mark Lett1992;3:289–96 (July).Bone PF, Scholder, EP. Olfaction and marketing: does it make sense to usescents? Working Paper, Department of Marketing, West Virginia Uni-versity, 1994.Bone PF, Scholder EP. Does it matter if it smells? Olfactory stimuli asadvertising executional cues. J Advertising 1998;27(4):29–40 (Winter).Booms BH, Bitner MJ. Marketing services by managing the environment.Cornell Hotel Restaurant Adm Q 1982;23:35–9 (May).Buck L, Axel R. A novel multigene family may encode odorant receptors: amolecular basis for odor recognition. Cell 1991;65:175–87 (April).Butcher D. Aromatherapy — its past and future. Drug Cosmet Ind 1998a;162(3):22–4 (March).Butcher D. Conditioning indoor environments using aroma technology.Drug Cosmet Ind 1998b (March1).Cann A, Ross DA. Olfactory stimuli as context cue in human memory. AmJ Psychol 1989;102:91–102 (Spring).Chebat JC, Vaillant D, Gelinas-Chebat C. Store music background in aselling context. J Bus Res 2001;54(1).Crowley AE. The two-dimensional impact of color on shopping. Mark Lett1993;4:59–69 (January).Darden WR, Babin BJ. Exploring the concept of affective quality: ex-panding the concept of retail personality. J Of Bus Res 1994;29(2):101–10 (February).Darden WR, Erdem O, Darden KK. A comparison and test of three causalmodels of patronage intentions. In: Darden WR, Lusch RF, editors.Patronage behavior and retail management. New York: North-Holland,1983 (491 p.).Davis RG. The role of nonolfactory context cues in odor identification.Percept Psychophys 1981;30(1):83–9.Dawson S, Bloch P, Ridgway NM. Shopping motives, emotional states andretail outcomes. J Retailing 1990;22:408–27 (Winter).Donovan R, Rossiter J. Store atmosphere: an environmental psychologyapproach. J Retailing 1982;58:34–57 (Spring).Dube´ L, Chebat JC, Morin S. The effects of background music on consum-ers’ desire to affiliate in buyer–seller interactions. Psychol Mark 1995;12(4):305–19 (July).Dunn C. The fragrance foundation looks ahead to spring. Household PersProd Ind 1997, vol. 40 (January 1).Ehrlichman H, Halpern JN. Affect and memory: effects of pleasant andunpleasant odors on retrieval of happy and unhappy memories. J PersSoc Psychol 1988;55(5):769–79.Ellen PS, Bone PF. Does it matter if it smells? Olfactory stimuly as adver-tising executional cues. J Advertising 1998;27(12–22):29.Engen T. The perception of odors. New York: Academic Press, 1982.Eroglu S, Harrel GD. Retail crowding: theoretical and strategic implica-tions. J Retailing 1986;2(4):346–63 (Winter).Eroglu S, Machleit KA. An empirical study of retail crowding: antecedentsand consequences. J Retailing 1990;66(2):201–21.Eroglu S, Machleit KA, Kellaris JJ. The impact of retail density and atmos-pheric music on retail shoppers’ crowding perceptions and store satis-faction. Presentation to Academy of Marketing Science, May 28, 1998,Norfolk, VA.Fisher JD. Situation-specific variables as determinants of perceived envi-ronmental aesthetic quality and perceived crowdedness. J Res Pers1974;8:177–88 (August).Garder MP, Siomkos GJ. Toward a methodology of assessing effects of in-store atmosphere. In: Lutz R, editor. Advances in consumer research,vol. 13. Provo (UT): Association for Consumer Research, 1985. p.27–31.Golden LG, Zimmerman DA. Effective retailing. Boston: HoughtonMifflin, 1986.Gulas CS, Bloch PH. Right under our noses: ambient scent and consumerresponses. J Bus Psychol 1995;10(1):87–98 (Fall).Harrell GD, Hutt MD, Anderson JC. Path analysis of buyer behavior underconditions of crowding. J Mark Res 1980;17:45–51 (February).Hui MK, Bateson JEG. Perceived control and the effects of crowding andJ.-C. Chebat, R. Michon / Journal of Business Research 56 (2003) 529–539538
  11. 11. consumer choice on the service experience. J Consum Res 1991;18:174–84 (September).Hunter BT. The sales appeal of scents (using synthetic food scents toincrease sales). Consum Res Mag 1995;78(10):8–10 (Oct).Izard CE, Kagan J, Zajonc RB, et al. Emotions, cognition, and behavior.Cambridge: Cambridge Univ. Press, 1984.Kirk-Smith MD. Culture and olfactory communication. In: Gardner RA, et al,editor. Ethological roots of culture. Boston: Kluwer, 1994. p. 385–406.Knasko S. Ambient odor’s effect on creativity, mood and perceived health.Chem Senses 1992;17(1):27–35.Knasko S, Gilbert AN, Sabini J. Emotional state, physical well-being andperformance in the presence of feigned ambient odor. J Appl Psychol1990;20(16):1345–57.Kotler P. Atmospherics as a marketing tool. J Retailing 1973;49:48–64 (winter).Laird DE. How the consumers estimate quality by subconscious sensoryimpressions: with special reference to the role of smell. J Appl Psychol1932;16:241–6 (June).Lazarus RS. Emotion and adaptation. New York: Oxford Univ. Press, 1991.Lorig TS, Schwartz GE. Brain and odor: alteration of human EEG by odoradministration. Psychobiology 1988;16(3):281–4.Mehrabian A, Russel JA. An approach to environmental psychology. Cam-bridge (MA): The MIT Press, 1974.Milliman RE. Using background music to affect the behavior of super-market shoppers. J Mark 1982;46(2):86–91.Milliman RE. The influence of background music on the behavior of res-taurant patrons. J Consum Res 1986;13:286–9 (September).Mitchell DJ, Kahn BE, Knasko SC. There’s something in the air: effects ofambient odor on consumer decision making. J Consum Res 1995;22:229–38 (September).Morrin M, Ratneshwar S. The impact of ambient scent on evaluation,attention and memory for familiar and unfamiliar brands. J Bus Res2000;49:157–65 (August).Moskowitz HR. Mind, body, and pleasure: an analysis of factors whichinfluence sensory hedonics. In: Kroeze JHA, editor. Preference behav-iour and chemoreception. London: Information Retrieval, 1979. p.131–47.Obermiller C, Bitner MJ. Store atmosphere: a peripheral cue for productevaluation. In: Stewart DC, editor. American psychological associationannual conference proceedings, consumer, psychology division. Wash-ington, DC: American Psychological Association, 1984. p. 52–3.Rappoport A. The meaning of the built environment. Beverly Hills (CA):Sage Publication, 1982.Richardson JTE, Zucco GM. Cognition and olfaction: a review. PsycholBull 1989;105(3):352–60.Russell JA, Pratt G. A description of the affective quality attributed toenvironments. J Pers Soc Psychol 1980;38(2):311–22.Schab FR. Odor memory: taking stock. Psychol Bull 1991;109(2):242–51.Schiffman SS. Preference: a multidimensional concept. In: Kroeze JHA,editor. Preference behaviour and chemoreception. London: InformationRetrieval, 1979. p. 63–81.Shostack GL. Breaking free from product marketing. J Mark 1977;41:73–80 (April).Smith P, Curnow R. Arousal hypothesis and the effects of music on pur-chasing behavior. J Appl Psychol 1966;50:255–6 (July).Spangenberg ER, Crowley AE, Henderson PW. Improving the store envi-ronment: do olfactory cues affect evaluations and behaviors? J Mark1996;60:67–80 (April).Stanley TJ, Sewall MA. Image inputs to a probabilistic model: predictingretail potential. J Mark 1976;40:48–53 (July).Turley LW, Milliman RE. Atmospheric effects on shopping behavior:a review of the experimental evidence. J Bus Res 2000;49:193–211 (August).Wakefield KL, Baker J. Excitement at the mall: determinants and effects ofshopping response. J Retailing 1998;74(4):515–39.Wheatly J, Chiu JSY. The effects of price, store image, and product andrespondent characteristics on perceptions of quality. J Mark Res 1977;14:81–186 (May).Yalch R, Spangenberg E. Effects of store music on shopping behavior. JConsum Mark 1990;7(2):55–63 (Spring).Zajonc RB, Markus H. Affect and cognition: the hard interface. In: IzardCE, Kagan J, Zajonc RB, editors. Emotions, cognition, and behavior.Cambridge: Cambridge Univ. Press, 1984. p. 63–103.J.-C. Chebat, R. Michon / Journal of Business Research 56 (2003) 529–539 539